CAREER: Towards Fault-tolerant Edge Computing for Cyber-Physical Systems: Distributed Primitives for Coordination under Cyber Attacks

边缘计算是一种新兴范例，它通过在用户处或附近分配一系列（边缘）服务器来提供必要的网络、存储和计算服务，从而扩展云计算。网络物理系统 (CPS) 将物理设备连接到互联网并相互连接。本研究探讨了边缘计算与 CPS（即基于边缘的 CPS）之间的结合。它吸引了学术界和工业界的极大关注，因为它有望从根本上改变人们与工程系统交互的方式。文献中仍然缺少一个基本要素——容错分布式原语的研究。这项研究的重点是分布式共识原语，它允许节点集合在存在各种故障和网络攻击的情况下作为一个连贯的实体工作。具体而言，存在三个相互关联的任务：（i）确定基于边缘的 CPS 的基本属性、限制和权衡； (ii) 设计能够自动适应不断变化的条件和不同网络攻击的算法，以保持良好的性能； (iii) 设计和实现一个真实的模拟器来研究实际场景下的表现。预期成果将为基于边缘的 CPS 提供高效且容错的服务，从而在 CPS 应用中实现更多创新。这项研究预计将通过确定一套设计分布式容错原语的原则、容错工具和分析来推进知识水平。这项工作将强调允许实际实施的模型和原语，并计划实际演示其容错性和性能优势。边缘计算和 CPS 预计将成为下一代基础设施的关键组成部分，因此使基于边缘的 CPS 具有容错能力可以产生大规模的效益。相反，由于缺乏基础理解而错失改进机会可能会产生严重的性能和容错后果。这项研究通过确定权衡和限制来帮助解决这些问题。拟议的研究还计划开发关于分布式系统的新教学材料，PI 计划通过波士顿学院的课程和向公众提供的在线教程来教授这些材料。此外，研究和推广工作将通过提供原则来帮助工业界和学术界设计基于边缘的 CPS 的容错原语，从而造福社会；加强学术界和工业界之间的伙伴关系；让妇女、残疾人和代表性不足的群体参与 STEM；该奖项反映了 NSF 的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

Edge computing is an emerging paradigm that extends cloud computing by allocating a range of (edge) servers at or near the user to provide necessary networking, storage, and computing services. Cyber-physical systems (CPS) connect physical devices to the Internet and to each other. This research explores the marriage between edge computing and CPS, namely edge-based CPS. It has attracted a tremendous amount of attention from both academia and industry because it is expected to fundamentally change the way people interact with engineered systems. One essential element –the study of fault-tolerant distributed primitives– is still missing in the literature. This research is focused on distributed consensus primitives that allow a collection of nodes to work as a coherent entity in the presence of various failures and cyberattacks. Concretely, there are three interrelated tasks: (i) identifying fundamental properties, limitations, and trade-offs in the context of edge-based CPS; (ii) designing algorithms that automatically adapt under changing conditions and varying cyberattacks to maintain good performance; and (iii) designing and implementing a realistic simulator to study performance under practical scenarios. The expected outcomes will enable efficient and fault-tolerant services for edge-based CPS, allowing more innovation in CPS applications. This research is expected to advance the state of knowledge by identifying a set of principles, fault-tolerance tools, and analyses for designing distributed fault-tolerant primitives. The work will emphasize models and primitives that allow for practical implementation, with plans for realistic demonstrations of their fault-tolerance and performance advantages. Edge computing and CPS are expected to be key components of next-generation infrastructure, so making edge-based CPS fault-tolerant can yield large-scale benefits. Conversely, missed opportunities for improvements caused by a lack of foundational understanding could have significant performance and fault-tolerance consequences. This research helps address these issues by identifying trade-offs and limitations. The proposed research also plans to develop new pedagogical material on distributed systems that the PI plans to teach, both through courses at Boston College and in online tutorials made available to the public. In addition, the research and outreach work will benefit society by providing principles to help industry and academia design fault-tolerant primitives for edge-based CPS; increasing partnerships between academia and industry; involving women, persons with disabilities, and underrepresented groups in STEM; and increasing public scientific literacy and public engagement with science and technology.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.